Annual reminder that the Current Population Survey can't be used to measure gaps in educational attainment between those who grew up rich and poor
brookings.edu/research/how-c…
brookings.edu/research/how-c…
The CPS is awesome for tracking education of the entire population.
Yay for high-quality, cross-sectional, government statistics! 2/N
Yay for high-quality, cross-sectional, government statistics! 2/N
The CPS is NO GOOD at telling us how adults' educational attainment differs by childhood socioeconomic status. 3/N
CPS is not a longitudinal datset. We don't track ppl from childhood to adulthood in the CPS.
When do we know someone's parents' income in CPS? When they live with their parents! 4/N
When do we know someone's parents' income in CPS? When they live with their parents! 4/N
The set of adults who live with their parents is, um, not representative of the population as a whole. 5/N
If you live away from your parents but are in college, you are typically considered to be in your parents' hhold for CPS survey purposes.
6/N
6/N
Consider two CPS hholds whose heads each have a 24-year-old kid
Bob lives away from parents, works --> not in CPS hhold
Sue lives away from parents, in college --> in CPS hhold
7/N
Bob lives away from parents, works --> not in CPS hhold
Sue lives away from parents, in college --> in CPS hhold
7/N
Hmmm this may explain why adult kids from richer families are more likely to appear on their parents' CPS records 8/N
🚨 BIAS SIREN 🚨
🚨 BIAS SIREN 🚨
If you try to use CPS to track educational attainment by parental income you will get batshit* results
*Whoops I spelled biased wrong 9/N
*Whoops I spelled biased wrong 9/N
After our post, the Pell Institute retracted its crazy 2015 estimate that 99% of kids in top income quartile earned a BA
But they're still at it. 10/N
But they're still at it. 10/N
The Pell Institute's annual report contains lots of great info, including stats from longitudinal surveys that provide CORRECT estimates of SES gaps in BA attainment 11/N
But the Pell Institute undermines the credibility of its entire report by leading - on the cover! - with unreliable, biased estimates. 12/N
There are reliable data showing enormous income gaps in BA attainment. It's a disservice to cloud the convo by using the wrong data to generate biased estimates. 13/N
Consider yourself innoculated against biased estimates when you see them. Very soon. 13/13
